On the Correlation of Tracer Diffusion Coefficients - American

School of Chemistry, University College, University of New South Wales, Australian Defence Force Academy,. Canberra, ACT 2600, Australia. A direct met...
0 downloads 0 Views 132KB Size
J. Chem. Eng. Data 1996, 41, 891-894

891

On the Correlation of Tracer Diffusion Coefficients Kenneth R. Harris* School of Chemistry, University College, University of New South Wales, Australian Defence Force Academy, Canberra, ACT 2600, Australia

A direct method is described for the correlation and comparison of sets of tracer (or limiting mutual) diffusion coefficients obtained under different conditions of temperature and density, which may or may not overlap, provided that at least one set has been obtained over a range of densities and temperatures. Recommendations are made for sets of diffusion data suitable for the calibration and cross-checking of high-pressure diffusion apparatus, particularly the Taylor dispersion (or chromatographic peakbroadening) apparatus.

Introduction In a recent paper, Funazukuri et al. (1994) reported binary diffusion coefficients of a series of organic compounds in hexane and other solvents. They used the Taylor dispersion technique at moderate pressures. As a check on accuracy, a comparison was made between measurements for benzene and toluene diffusing in hexane and literature data for tracer diffusion coefficients for these systems obtained with the diaphragm-cell method. These diaphragm-cell results were obtained with a high-pressure apparatus for a single isotherm for each system at 298 K to 400 MPa (Dymond and Woolf, 1982), supplemented at atmospheric pressure by higher precision glass diaphragmcell results again at 298 K and Taylor results between 273 K and 333 K (Dymond, 1981). The high-pressure Taylor dispersion measurements were made between 303 K and 333 K and at 16 MPa and 25 MPa. Funazukuri et al. say, “ Since these measurement conditions are different, the D12 values cannot be compared directly. Thus, the D12 data are compared in plots of T/(ηD12) vs solvent molar volume.” This indirect procedure combines errors in four experimental quantities in attempting to check one. However a direct method does exist for the comparison of sets of diffusion coefficients (and other transport properties such as the viscosity) obtained under different conditions, which may or may not overlap, provided that at least one set has been obtained over a range of densities and temperatures (Harris, 1982, 1993, 1995). Here, the method is adapted to the case of tracer diffusion and applied to the results of Funazukuri et al. (1994) as an example. More importantly, it provides a basis for the calibration and cross-checking of the Taylor dispersion apparatus operated above atmospheric pressure using systems studied by the more established diaphragm-cell method. (Suitable systems for atmospheric pressure calibration, for which precise data are more abundant, have been discussed elsewhere (Harris, 1991).) Method Regularities in the dependence of transport properties on molar volume have long been known, dating back more than 80 years to the viscosity studies of Batschinskii (1913). In particular, the fluidity (Φ) and self-diffusion coefficient (D) for simple, non-hydrogen-bonded molecular liquids are almost linear functions of the molar volume at moderate pressures, as are the same properties for the hard sphere * E-mail: [email protected].

S0021-9568(96)00066-0 CCC: $12.00

fluid as obtained by computer simulation (Dymond, 1974a). At high pressures, D-V (Parkhurst and Jonas, 1975) and Φ-V (Van Wijk and Seeder, 1937) isotherms curve away from a straight line dependence, real liquids being more fluid than the hard sphere model would predict. This curvature is conveniently reproduced by the equation (Harris, 1982)

D* ) ζ1 + ζ2Vref/(1 + ζ3/Vref)

(1)

D* is a reduced diffusion coefficient introduced by Dymond (1974b) as an aid in the application of the hard sphere model to real fluids. Vref is a reference molar volume (see below) and the ζi are fitted coefficients. For tracer diffusion D* is given by

D* T2 ≡

nDT2

( )

V V (nDT2) 0 ∞

2/3

(2)

(using the notation DT2 to indicate the limiting value of the mutual diffusion coefficient D12 for the diffusion of a trace of solute 2 in solvent 1). n is the number density, and (nDT2)∞ is the density-diffusion coefficient product for a dilute gas of hard spheres given by the ChapmanEnskog equation in its first (composition independent) approximation (Chapman and Cowling, 1960),

(nDT2)∞ )

3 kT 8σ122 2πµ

1/2

( )

(3)

In this equation, µ is the reduced molecular mass (m1m2/ (m1 + m2)), k is Boltzmann’s constant, and σ12 is the mean diameter of the two species. V0 in eq 2 is σ123/x2. Inspection of these equations shows DT2* to be independent of σ12, so it is a function of only the molecular masses, temperature and density. For many fluids, the reduced diffusion coefficient isotherms on a DT2*-V plot are similar in the geometric sense and may be superposed onto a single reference isotherm, Tref, chosen arbitrarily, by the coordinate transformation

Vref ) V(1 - ξ1(T - Tref) - ξ2(T - Tref)2)

(4)

Self-diffusion data for hexane (Harris, 1982), so reduced, are shown in Figure 1 as an example. Equation 2 simplifies to © 1996 American Chemical Society

892 Journal of Chemical and Engineering Data, Vol. 41, No. 4, 1996 Table 1. Coefficients of Eqs 1 and 4 for the Tracer Diffusion of Benzene and Toluene in Hexane (Diaphragm Cell Data) ζ1 102ζ2/mol cm-3 10-2ζ3/cm3 mol-1 103ξ1/K-1 105ξ2/K-2 Tref/K stand devn/% expt acc/%

benzene + hexane

toluene + hexane

-1.677 06 0.942 285 -0.434 897 -0.767 240 0.557 835 273.15 2.5 (2-4

-1.869 28 1.011 48 -0.458 347 -0.612 878 -0.652 308 273.15 3.1 (2-4

Table 2. Comparison of the Taylor Dispersion Results for the Tracer Diffusion of Benzene and Toluene in Hexane of Funazukuri et al. with the Correlation Figure 1. Reduced self-diffusion coefficients for hexane [(b) 223 K; (0) 248 K; ([) 273 K; (3) 298 K; (2) 333 K], shifted to a common reference isotherm, 298 K (solid line).

xTµ

DT2 D*T2 ) aD 1/3 (V)

(5)

with aD ) 17.44 J1/2 K1/2 mol2/3 when DT2, V, µ, and T have units of 10-9 m2/s, cm3/mol, kg/mol, and K, respectively. Using these equations DT2(T,V) data can be fitted by nonlinear least squares yielding five parameters, ζi (i ) 1-3) and ξj (j ) 1, 2): eqs 1 and 4 can then be used to interpolate DT2(T,V) within the temperature and volume range of the original results. This approach has been used successfully for self-diffusion and viscosity data of a number of molecular liquids, including aromatic and aliphatic hydrocarbons (Harris, 1982, 1993; Harris et al., 1993) and substituted methanes (Harris, 1993; Harris et al., 1990). It has also formed a basis for the calculation of joint fits (Harris et al. 1993) for these two properties using a common Vref function as well as the correlation of the transport properties for several series of related compounds through the construction of “family” curves from data for representative substances (Harris, 1993, 1995). Results The results of the fitting procedure for benzene and toluene in hexane using the data sets of Dymond (1981) and Dymond and Woolf (1982) are given in Table 1. A comparison of the results of Funazukuri et al. (1994) with values predicted by the correlations for these two systems is given in Table 2. Molar volumes were taken from the equation of state previously used for hexane (Harris, 1982). Generally, the difference between the data sets for the two systems is greater than the sum of the experimental errors. A comparison has also been made between the DymondWoolf diaphragm-cell tracer diffusion results and the selfdiffusion data for hexane referred to above, which were obtained by the nmr spin-echo technique. Within experimental error, the tracer diffusion coefficients have the same volume dependence as the self-diffusion coefficient, which is consistent with the lack of specific solute-solvent interactions in these systems. This result was obtained by fitting the tracer data to the equations (Harris, 1995)

D*T2 ) RD* S1

(6a)

Vref,T ) rVref,S

(6b)

where the subscripts T and S indicate tracer and selfdiffusion, respectively. For benzene in hexane, R ) 1.13 ( 0.02 and r ) 0.999 ( 0.002, with a standard deviation

T/K

p/ MPa

V/cm3 mol-1

D/10-9 102(D*expt - D*calc)/ D* expt m2 s-1 D*expt D*calc

303.2 313.2 313.2 323.2 323.2

16 16 25 16 25

129.36 130.97 129.36 132.65 134.40

benzene + hexane 3.89 0.156 0.176 4.48 0.176 0.191 3.91 0.154 0.179 4.86 0.187 0.206 5.27 0.202 0.219

-12.7 -8.6 -16.2 -10.2 -8.6

303.2 313.2 313.2 323.2 323.2

16 16 25 16 25

129.36 130.97 129.36 132.65 134.40

toluene + hexane 3.79 0.158 0.169 4.32 0.177 0.182 3.82 0.157 0.170 4.76 0.191 0.195 5.24 0.209 0.208

-6.6 -2.9 -8.4 -2.0 0.6

of 3.5% and for benzene in toluene, R ) 1.08 ( 0.03 and r ) 0.999 ( 0.003, with a standard deviation of 3.6%. This agreement between the volume dependence of the tracer and self-diffusion coefficients, obtained by different experimental techniques, allows one some confidence in the Dymond-Woolf tracer results. The systematic difference of the results of Funazukuri et al. may possibly be due to their use of a spectroscopic grade of hexane. Some manufacturers do not remove branched alkanes from this grade and the purity can be suprisingly low. For the benzene system, rather high concentrations were injected (10 to 50% by volume). Though the injected solution is considerably diluted in the Taylor technique, the result that the measured diffusion coefficients were independent of concentration under these conditions is suprising given the strong composition dependence of the mutual diffusion coefficient in this system at atmospheric pressure (Harris et al., 1970). Whatever the reason for the differences between these particular data sets, the important point is that the method described here offers a general, direct way of making comparisons and analysis. Nevertheless, this comparison would be fairer if more than one high-pressure isotherm were available for the reference data systems. At present, with this lack of data, (benzene + hexane) and (toluene + hexane) are not ideal for the calibration and cross-checking of the high-pressure Taylor dispersion diffusion apparatus, and alternatives are required. The literature on high-pressure diffusion in liquids has been examined and the system (acetonitrile + methanol) is a better candidate for a calibration system. It has been more extensively studied by the high-pressure diaphragmcell method than has (benzene + hexane) or (toluene + hexane), with data available for this system over the whole composition range and between 283 K and 313 K at pressures to 260 MPa (Hurle and Woolf, 1982b). Since acetonitrile and methanol self-diffusion data obtained with the same apparatus have been cross-checked against nmr spin-echo results (Hurle and Woolf, 1982a; Hurle et al., 1985), one can have some confidence in these tracer results.

Journal of Chemical and Engineering Data, Vol. 41, No. 4, 1996 893 Table 4. Equations of State for Acetonitrile and Methanol 10-3R00/MPa 10-6R01/MPa K 10-1R10 10-3R11/K 104R20/(MPa)-1 R21/K (MPa)-1 std devn in κ/MPa std devn in V/%a β0/cm3 mol-1 103β1/cm3 mol-1 K-1 105β2/cm3 mol-1 K-2 std devn/cm3 mol-1

Figure 2. Reduced tracer diffusion coefficients for acetonitrile in methanol [(b) 283 K; (9) 298 K; (2) 313 K] and methanol in acetonitrile [(O) 283 K; (0) 298 K; (4) 313 K] with Tref ) 273.15 K.

CH3CN

CH3OH

-1.251 65 0.646 241 0.437 740 -0.121 637 -79.929 8 2.006 38 8.7 0.1 43.829 3b -12.752 9 14.441 0 0.006

-0.619 333 0.426 059 0.229 812 0.110 283 -0.464 340 -0.812 999 5.4 0.06 32.579 5 5.659 4 7.277 0

a Data for the fits were generated from smoothed functions for particular isotherms reported by Easteal and Woolf (1985a, 1985b) (278 < T/K < 323, p < 280 MPa), so the fit is better than would have been obtained from the raw experimental data. The advantage of the Hayward equation used is that it allows easy interpolation between experimental p-V isotherms, as was required here. b Atmospheric pressure data for CH CN from Handa and Benson 3 (1981), French (1987), and Sakurai (1992); β coefficients from Easteal and Woolf (1985b) for CH3OH.

must be careful to avoid convergence on false minima on the multidimensional parameter-residual surface. This is dependent on using good initial estimates for the fitted parameters, though these can be found by trial and error with little difficulty. Experience with a variety of systems has shown that ζ2 should be positive and ζ3 negative and about a third to a half of the mean molar volume of the data set in magnitude. Fortran programs for fitting data are available from the author. Appendix

Figure 3. Deviation plot for acetonitrile in methanol [(b) 283 K; (9) 298 K; (2) 313 K] and methanol in acetonitrile [(O) 283 K; (0) 298 K; (4) 313 K]. Table 3. Coefficients of Eqs1 and 4 for the System Methanol + Acetonitrile

ζ1 10ζ2/mol cm-3 10-2ζ3/cm3 mol-1 102ξ1/K-1 105ξ2/K-2 Tref/K stand devn/% expt acc/%

DT(CH3OH) in CH3CN

DT(CH3CN) in CH3OH

-1.380 15 0.186 377 -0.189 567 -0.142 809 0.882 390 273.15 2.4 (2

-0.989 654 0.174 065 -0.145 398 -0.175 025 -0.553 806 273.15 1.5 (2

With suitable detectors, either the data set for DT(acetonitrile) in methanol or that for DT(methanol) in acetonitrile can be used for calibration and cross-checking. As both these substances show strong absorption only in the low ultraviolet, so that spectrophotometric detectors would have to be set on a band shoulder rather than a peak, the data are perhaps better suited to the calibration of bulk property detectors than to a uv spectrophotometer, unless the latter is very stable with respect to drift. Table 3 includes coefficients for eqs 1 and 4 for these two systems. The data are plotted in Figure 2 as reduced diffusion coefficients against reference molar volumes, and deviations from the fit are shown in Figure 3. The fit is satisfactory for both systems, though that for the diffusion of methanol in acetonitrile is the better of the two. It should be noted that as the fitting is a nonlinear least squares procedure (adapted from Wentworth, 1965), one

Table 4 summarizes good quality literature pVT data for acetonitrile (Easteal and Woolf, 1985a) and methanol (Easteal and Woolf, 1985b) which can be used to obtain molar volumes. A Hayward equation of state is used, where the linear secant modulus

κ ) Vσ(p- pσ)/(Vσ - V)

(7)

is expressed as a polynomial in the pressure, p, 4

κ)

∑R p

i

(8)

i

i)0

with temperature dependent coefficients

Ri ) R0i + R1i/T

(9)

Vσ is the liquid molar volume at atmospheric pressure or along the saturation curve. These Vσ values were fitted to a polynomial in the temperature, 2

Vσ )

∑β T

i

i

(10)

i)0

The coefficients Rij and βk are given in Table 4. Acknowledgment The author is grateful for referees’ comments and the detection of an error, now happily corrected. Literature Cited Batschinskii, A. I. Inner Friction of Liquids. Z. Phys. Chem. 1913, 84, 643-706.

894 Journal of Chemical and Engineering Data, Vol. 41, No. 4, 1996 Chapman, S.; Cowling, T. G. The Mathematical Theory of Non-Uniform Gases; University Press: Cambridge, 1960; p 250. Dymond, J. H. Corrected Enskog Theory and the Transport Coefficients of Liquids. J. Chem. Phys. 1974a, 60, 969-973. Dymond, J. H. The Interpretation of Transport Coefficients on the Basis of the Van der Waals Model. Physica 1974b, 75, 100-114. Dymond, J. H. Limiting Diffusion in Binary Nonelectrolyte Mixtures. J. Phys. Chem. 1981, 85, 3291-3294. Dymond, J. H.; Woolf, L. A. Tracer Diffusion of Organic Solutes in n-Hexane at Pressures up to 400 MPa. J. Chem. Soc., Faraday Trans. 1 1982, 78, 991-1000. Easteal, A. J.; Woolf, L. A. p,V,T and Derived Thermodynamic Data for Toluene, Trichloromethane, Dichloromethane, Acetonitrile, Aniline and n-Dodecane. Int. J. Thermophys. 1985a, 6, 331-351. Easteal, A. J.; Woolf, L. A. (p,Vm,T,x) Measurements for {(1-x) H2O + x CH3OH} in the Range 278 to 323 K and 0.1 to 280 MPa. 1. Experimental Results, Isothermal Compressibilities, Thermal Expansivities and Partial Molar Volumes. J. Chem. Thermodynam. 1985b, 17, 49-62. French, H. T. Vapour Pressures and Activity Coefficients of (Acetonitrile + Water) at 308.15 K. J. Chem. Thermodynam. 1987, 19, 11551161. Funazukuri, T.; Nishimoto, N.; Wakao, N. Binary Diffusion Coefficients of Organic Compounds in Hexane, Dodecane, and Cyclohexane at 303.2-333.2 K and 16 MPa. J. Chem. Eng. Data 1994, 39, 911915. Handa, Y. P.; Benson, G. C. Thermodynamics of Aqueous Mixtures of Nonelectrolytes IV. J. Solution Chem. 1981, 10, 291-300. Harris, K. R. Temperature and Density Dependence of the Selfdiffusion Coefficients of n-Hexane from 223 to 333 K and up to 400 MPa. J. Chem. Soc., Faraday Trans. 1 1982, 78, 2265-2275. Harris, K. R. On the Use of the Edgeworth-Crame´r Series to Obtain Diffusion Coefficients from Taylor Dispersion Peaks. J. Solution Chem. 1991, 20, 595-606. Harris, K. R. Correlation of Dense Fluid Self-Diffusion and Shear Viscosity Coefficients. High Temp. High Pressures 1993, 25, 359366. Harris, K. R. Correlation of Dense Fluid Self-Diffusion, Shear Viscosity, and Thermal Conductivity Coefficients. Int. J. Thermophys. 1995, 16, 155-165.

Harris, K. R.; Pua, C. K. N.; Dunlop, P. J. Mutual and Tracer Diffusion Coefficients and Frictional Coefficients for the Systems BenzeneChlorobenzene, Benzene-n-Hexane, and Benzene-n-Heptane at 25 °C. J. Phys. Chem. 1970, 74, 3518-3529. Harris, K. R.; Lam, H. N.; Raedt, E.; Easteal, A. J.; Price, W. E.; Woolf, L. A. The Temperature and Density Dependences of the Selfdiffusion Coefficient and Shear Viscosity of Liquid Trichloromethane. Mol. Phys. 1990, 71, 1205-1221. Harris, K. R.; Alexander, J. J.; Goscinska, T.; Malhotra, R.; Woolf, L. A.; Dymond, J. H. Temperature and Density Dependences of the Self-diffusion Coefficients of n-Octane and Toluene. Mol. Phys. 1993, 78, 235-248. Hurle, R. L.; Woolf, L. A. Self-diffusion in Liquid Acetonitrile under Pressure. J. Chem. Soc., Faraday Trans. 1 1982a, 78, 2233-2238. Hurle, R. L.; Woolf, L. A. Tracer Diffusion in Methanol and Acetonitrile under Pressure. J. Chem. Soc., Faraday Trans. 1 1982b, 78, 29212928. Hurle, R. L.; Easteal, A. J.; Woolf, L. A. Self-diffusion in Monohydric Alcohols under Pressure. J. Chem. Soc., Faraday Trans. 1 1985, 81, 769-779. Parkhurst, H. J., Jr.; Jonas, J. Dense Liquids. 1. The Effect of Density and Temperature on Self-diffusion of Tetramethylsilane and Benzened6. J. Chem. Phys. 1975, 63, 2698-2704. Sakurai, M. Partial Molar Volumes for Acetonitrile + Water. J. Chem. Eng. Data 1992, 37, 358-362. Van Wijk, W. R.; Seeder, W. A. The Influence of Temperature and the Specific Volume on the Viscosity of Liquids. Physica 1937, 4, 10731088. Wentworth, W. E. Rigorous Least Squares Adjustment. Application to Some Non-linear Equations, I and II. J. Chem. Educ., 1965, 42, 96-103; 162-167.

Received for review February 14, 1996. Accepted April 30, 1996.X

JE9600663 X

Abstract published in Advance ACS Abstracts, June 15, 1996.